Selective Action of 2’,3’-Didehydro=2’,3’-dideoxythymidine Triphosphate on Human Immunodeficiency Virus Reverse Transcriptase and Human DNA Polymerases*

This study used DNA primer extension and sequencing gel analyses to evaluate the molecular action of 2’,3’-didehydro-2’,3’-dideoxythymidine triphosphate (D4TTP),’ in comparison with 3’-azido-2’,3’-dideox-ythymidine triphosphate (AZTTP), on DNA strand elongation by human immunodeficiency virus reverse transcriptases (HIV-RT) and human DNA polymerases a (pol a) and c (pol e) purified from T-lymphoblastoid CEM cells. D4TTP was preferentially incorporated into the T sites of the elongating DNA strand by HIV- RT and terminated DNA synthesis at the incorporation sites. The DNA chain termination activity of DITTP was equipotent to that of AZTTP. In contrast, D4TTP was a poor substrate for pol a and pol t. The analogue was incorporated into DNA by the human enzymes about 10,000- to 20,000-fold less efficiently than by HIV-RT, whereas the incorporation of AZTTP by pol a and pol c was not detectable by the DNA primer extension assay. Pol t, an enzyme with 3’ + B’-exonu-clease activity, was unable to remove the incorporated 2‘,3’-didehydro-2’,3’-dideoxythymidine monophos- phate (D4TMP) from the 3’-end of the DNA strand,


Selective Action of 2',3'-Didehydro=2',3'-dideoxythymidine Triphosphate on Human Immunodeficiency Virus Reverse Transcriptase and Human DNA Polymerases*
(Received for publication, July 29, 1991) Peng Huang, David Farquhar, and William PlunkettS This study used DNA primer extension and sequencing gel analyses to evaluate the molecular action of 2',3'-didehydro-2',3'-dideoxythymidine triphosphate (D4TTP),' in comparison with 3'-azido-2',3'-dideoxythymidine triphosphate (AZTTP), on DNA strand elongation by human immunodeficiency virus reverse transcriptases (HIV-RT) and human DNA polymerases a (pol a) and c (pol e) purified from T-lymphoblastoid CEM cells. D4TTP was preferentially incorporated into the T sites of the elongating DNA strand by HIV-RT and terminated DNA synthesis at the incorporation sites. The DNA chain termination activity of DITTP was equipotent to that of AZTTP. In contrast, D4TTP was a poor substrate for pol a and pol t. The analogue was incorporated into DNA by the human enzymes about 10,000-to 20,000-fold less efficiently than by HIV-RT, whereas the incorporation of AZTTP by pol a and pol c was not detectable by the DNA primer extension assay. Pol t, an enzyme with 3' + B'-exonuclease activity, was unable to remove the incorporated 2',3'-didehydro-2',3'-dideoxythymidine monophosphate (D4TMP) from the 3'-end of the DNA strand, whereas 3'-azido-2',3'-dideoxythymidine monophosphate was excised from DNA by pol t at about 20% of the rate for normal deoxynucleotide excision. The preferential incorporation of D4TTP by HIV-RT appears to be a molecular basis for the selective anti-HIV activity of D4T, whereas the inability of pol c to remove D4TMP from DNA may be related to the cytotoxicity of this compound.
The human immunodeficiency virus reverse transcriptase (HIV-RT)' is the enzyme responsible for catalyzing the conversion of viral RNA to double-stranded DNA, which is then integrated into the host cellular DNA (1). Because HIV-RT plays an essential role in the viral life cycle, this enzyme is * This work was supported by Grant A128213 from the National Institute of Allergy and Infectious Diseases and CA28596 from the National Cancer Institute, Department of Health and Human Services. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be sent: Dept. of Medical Oncology, Box 52, The University of Texas M. D. Anderson Cancer Center, Houston, TX 77030.
Studies of the metabolism and action of D4T have demonstrated that this compound enters the cell by nonfacilitated diffusion (22) and is phosphorylated by cellular enzymes to its mono-, di-, and triphosphates. Unlike AZT, D4T showed a low affinity to thymidine kinase ( K , = 140 p~) in several cell lines (19,20). Thus, the conversion to its monophosphate by this enzyme is the rate-limiting step in its metabolic activation pathway to triphosphate (D4TTP) (20,23). D4TTP has been shown to be a potent and selective inhibitor of HIV-RT with a Ki value of 12-32 nM when poly(rA):oligo(dT) was used as the template (13,14), whereas much greater concentrations of D4TTP are required to inhibit cellular pol a (19). However, the molecular mechanism of this selectivity remains to be elucidated. In this study, we employed a DNA primer extension technique and sequencing gel analyses to evaluate the molecular action of D4TTP, in comparison with AZTTP, on DNA strand elongation by HIV-RT and human DNA pol a and pol t purified from T-lymphoblastoid CEM cells. The ability of the 3' + 5'-exonuclease activity of pol t to remove the incorporated D4TMP and AZTMP molecules from DNA was also investigated.

EXPERIMENTAL PROCEDURES
Materials"D4TTP and AZTTP were chemically synthesized and purified according to the procedures described previously (13,24). The M13 dideoxy sequencing kit was purchased from Bethesda Research Laboratories. The M13 mp18(+) strand DNA, the 17-base

Selective Action
of D4TTP on HIV-RT noaffinity chromatography and contained two polypeptides (p66 and p51) when analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (25). Pol CY and pol c were purified from human Tlymphoblastoid CEM cells and characterized as previously described (12,26). DNA Primer Extension Assay-The 17-base oligonucleotide was labeled with ' ?P at its 5'-end and annealed to its complementary site on the single stranded M13 mp18 (+) DNA template as described previously (12, 2G). The "'P-labeled primer/template where the numbers indicate the nucleotide sites from the 3'-end of the primer, was precipitated with 3 volumes of ethanol, washed with cold ethanol, dissolved in distilled H20, and used as the template for DNA extension.
The primer extension reaction mixture (10 pl) contained 20 mM Tris-HCI (pH 7.5), 8 mM MgC12,0.5 mM dithiothreitol, 10 mM NaCI, 20 pg of bovine serum albumin/ml, 40 ng of labeled primer/template, 0.1 unit of pol CY or pol c, or 0.005 unit of HIV-RT, 30 pM each of dATP, dCTP, and dGTP, and various concentrations of dTTP and D4TTP or AZTTP, as indicated in the figure legends. The reactions were incubated at 37 "C for 30 min and analyzed by electrophoresis through a 10% polyacrylamide sequencing gel. The radioactivity in DNA bands on the sequencing gel was quantitated with a Betascope 603 blot analyzer (Betagen, Waltham, MA). The relative velocity of D4TTP incorporation was determined by dividing the radioactivity in the target site (site 8) by the radioactivity of the band 1 nucleotide shorter (site 7) as previously described (27,28).
The K,,, and V,,, values were then calculated from the Michaelis-Menten equation, using a computer-assisted program (29).
Excision of D4TMP and AZTMP from DNA-Oligomers with either d4TMP or AZTMP incorporated at the 3' terminus were constructed by incubating the "P-labeled 17-base primer/M13 mp18 DNA hybrid with 30 p~ each of dATP, dCTP, and dGTP, and 10 p~ of either D4TMP or AZTTP, and 0.01 unit of HIV-RT at 37 "C for 30 min. Either B4TTP or AZTTP was incorporated into site 8, resulting in an analogue-terminated oligomer of 25 nucleotides. The reaction products were denatured and separated by 15% polyacrylamide sequencing gel electrophoresis. The 25-base oligomer containing either D4TMP or AZTMP at it 3'-end was recovered from the gel and annealed to the M13 mp18 (+) DNA as previously described (26). As a control, the 17-base primer was also isolated from the gel and hybridized to the M13 mp18 (+) DNA template. These DNA hybrids were used as the substrates of pol c for the excision assay as previously described (26).

RESULTS
The DNA primer extension assay was used to evaluate the action of D4TTP on HIV reverse transcriptase and human pol a and pol e. In Fig. 1, lane 1 contains a reaction with HIV-RT, dATP, dCTP, and dGTP without dTTP (Treaction). The intense band at site 7 (1 nucleotide before the first T site) demonstrated that the majority of the 17-base primers were extended to and terminated at the site 1 nucleotide before the first T site, because there was no dTTP present in the reaction. The band at the first T site (site 8) indicated that some noncomplementary dNTP molecules were misincorporated into the T site by HIV-RT, confirming its errorprone properties (30-32). A small portion of the primers containing misincorporated nucleotide were further extended, resulting in the production of longer oligomers, as indicated by the two light bands at the sites 1 nucleotide before the third and fourth T sites.
When increasing concentrations of D4TTP were added to the Treaction, the HIV-RT was able to efficiently use the analogue as the substrate for incorporation into the first T site (site 8 ) (Fig. 1, lanes 2-5). This incorporation led to a decrease in the number of primers at the pre-T site (site 7) and an accumulation of primers at the first T site. The incorporated D4TMP molecules terminated the primer extension; no DNA bands were detected beyond the first T site.
In the reactions that contained 30 PM each of dATP, dCTP, In the experiments with 0.1 unit of purified human pol a and pol t, higher concentrations of D4TTP (10-30 PM) were required to force the enzymes to incorporate trace amounts of D4TTP into the T sites (Fig. 2, lunes 4,5, 14, and 15). The amount of HIV-RT, pol a, and pol e used in this experiment produced approximately same amount of high molecular weight DNA products in the absence of inhibitor, using the primed M13 mp18 DNA as the template. Quantitation of the radioactivity in this and another sequencing gel during a 100min counting period showed that at 10 PM D4TTP, 830 f 65 and 1172 f 209 counts were incorporated into the first T site in the reactions with pol (Y and pol e, respectively. In contrast, 4822 f 455 counts were detected at the same site by HIV-RT when the D4TTP concentration was only 1 PM (Fig. 1,  2). These results demonstrate the preferential incorporation of D4TTP into DNA by HIV-RT. Furthermore, in the reactions with pol a or pol t, 1 ~L M dTTP effectively competed with 30 p~ D4TTP to support the synthesis of high molecular weight DNA strands ( lanes 6 and 16), suggesting that dTTP is the preferred substrate for the human enzymes.
To quantitate the kinetics of D4TTP incorporation, the primed M13 mp18 (+) DNA was incubated with HIV-RT, pol a, or pol t, 100 p~ each of dATP, dCTP, and dGTP, and various concentrations of D4TTP (Fig. 3). D4TTP was incorporated into the first T site (site 8) by HIV-RT (lanes 1-7), pol a (lanes 8-14), and pol t (lanes 15-21), although the incorporation efficiencies differed among the enzymes. In lanes 15-21, the bands below the 17-base primer reflect the short DNA fragments, resulting from the digestion of the longer DNA oligomers by the 3' + 5'-exonuclease activity of pol e, The lower panel of Fig. 3 quantitatively illustrates the incorporation of D4TMP into the first T site (site 8) by the viral and human enzymes. At the D4TTP concentrations ranged from 0.02 to 1 p~, HIV-RT was able to incorporate substantial amounts of D4TMP into DNA in a concentrationdependent manner. A separate experiment indicated that at higher D4TTP concentrations (3-30 pM), the incorporation of D4TMP into site 8 by HIV-RT reached a plateau (data not shown). In contrast, only a small amount of the analogue was incorporated by either pol a or pol t at higher D4TTP concentrations (3-400 p~) .
The radioactivity in sites 7 and 8 of each lane was quantitated, and the incorporation kinetics ( K , and V,,,,,) were determined as described under "Experimental Procedures." As shown in Table I, the K , values for D4TTP incorporation into the first T site by HIV-RT was about 400-fold less than that for pol a or pol t, indicating that D4TTP has much higher affinity to HIV-RT than to pol a and pol t. The substrate efficiency (V,,,JK,,,) of HIV-RT for D4TTP was about 10,000-and 20,000-fold greater than that of pol a and pol t, respectively.  The action of D4TTP on HIV-RT and pol a and pol c was compared with that of AZTTP under identical conditions. As illustrated in Fig. 4, the incorporation of D4TTP and AZTTP into DNA by HIV-RT was similar under conditions measuring initial reaction rates over 30 min (lanes 2 and 3). Quantitation of two separate sequencing gels revealed that HIV-RT incorporated 22,912 f 1497 counts and 24,876 f 1498 counts into the first T site in reactions containing 30 p~ D4TTP and AZTTP, respectively (lanes 2 and 3 ) . A separate experiment demonstrated that the two analogues were incorporated by  13). Little effect was exhibited by either D4TTP or AZTTP on DNA primer elongation by pol a (lanes [10][11][12] or pol e (lanes [16][17][18] when dTTP was present in the reaction mixtures. To evaluate the ability of DNA pol t to remove the incorporated D4TMP and AZTMP from the 3'-end of the DNA with its 3' + 5'-exonuclease activity, the 25-base oligomers with either D4TMP of AZTMP at the 3'-ends were constructed and annealed to the M13 mp18 (+) DNA. These DNA complexes were used as the substrates for excision by pol e. The normal 17-base primer was processed through the same procedures and used as the control. In Fig. 5, the upper panel demonstrates that during a 50-min incubation, pol e removed a substantial amount of normal nucleotides from the 17-base primer, producing shorter oligomers (lanes [1][2][3][4][5]. In contrast, the enzyme was unable to excise D4TMP from the 3'-end of the 25-base primer (lanes [8][9][10][11][12]. When the primer with AZTMP at the 3'-end was used as the substrate, pol t removed small amounts of the analogue from DNA at a slow rate (lanes [15][16][17][18][19]. Quantitation of the radioactivity in the bands showed that AZTMP was removed at about 20% of the rate for normal nucleotide excision (Fig. 5, lower panel). It was expected that if the DNA chain-terminating molecule (D4TMP or AZTMP) was removed from the 3'-end of the DNA primer by pol t, HIV-RT would be able to extend the primer to a high molecular weight DNA strand in the presence of four dNTPs. As illustrated in lane 6 of Fig. 5, HIV-RT extended the normal 17-base primer to high molecular weight DNA in the presence of dNTPs. When both HIV-RT and pol e were included in the reaction containing the 17-base primer and four dNTPs, products of polymerization, as well as excision, were revealed ( h n e 7). The extension products appeared to be generated mainly by HIV-RT, because the natural pause band at about the 100-nucleotide site catalyzed by pol t (Fig.  4, lane 18) was absent in this reaction (Fig. 5, lane 7). In the presence of pol e and four dNTPs, however, HIV-RT was unable to extend the 25-base primer with 3'-D4TMP (lanes 13 and 14), indicating that the chain-terminating D4TMP residue still remained at the 3'-end of the primer after incubation with pol t. In contrast, pol e was able to excise a small amount of AZTMP from the 3'-end of the 25-base primer, which allowed the primer to be extended to high molecular weight DNA (lanes 20 and 21 ). Presumably, the excision products (24-base or smaller bands) were not detectable in those two reactions because the presence of four dNTPs enabled the extension of the short oligomers (<25 bases) to longer strands.

DISCUSSION
This study demonstrated that D4TTP was preferentially incorporated into DNA by HIV reverse transcriptase. In contrast, human DNA polymerases were unable to efficiently use D4TTP as the substrate for incorporation. Quantitative comparison showed that HIV-RT was about 10,000-to 20,000fold more efficient than pol a and pol e in incorporating D4TTP into DNA (Table  I). This selective incorporation resulted in termination of DNA synthesis by the viral enzyme. Incubation of CEM cells with 2 p~ D4T (23) or MT-4 cells with 1 p~ D4T (19) for 24 h resulted in the accumulation of 1.4 and 0.22 p~ cellular D4TTP, respectively. These concentrations are close to the K,,, value for D4TTP incorporation by HIV-RT (0.35 p~) and far below the K, values for human pol a (139.9 p~) and pol e (156.6 pM). Thus, it may be important to monitor the cellular pharmacology of D4T. A D4T administration schedule that maintains cellular D4TTP at levels that terminate viral DNA synthesis but does not affect cellular DNA replication would optimize the therapeutic response.
The selective action of D4TTP on HIV-RT is similar to that of AZTTP (12). The comparison of their molecular action on DNA primer extension by HIV-RT revealed that D4TTP and AZTTP were incorporated into the growing DNA strand at the same efficiency and terminated DNA synthesis in same manner. This is in agreement with the observation that D4TTP and AZTTP are equipotent in inhibiting HIV-RT (19). Both AZT and D4T are analogues of thymidine, and neither has a hydroxyl linked to the 3'-carbon of the sugar moiety. Their DNA chain termination activity is predicted because the 3'-hydroxyl is required for DNA strand elongation. The molecular mechanism of the selective action of these compounds on HIV-RT, however, is based on their preferential substrate efficiency for HIV-RT relative to human DNA polymerases. It appears that a thymidine analogue with a modification at the 3"position is likely to be a preferred substrate for HIV-RT. Thus, molecular evaluation of the interaction between nucleotides bearing modifications at the 3"position and HIV-RT and human DNA polymerases may be useful in the design of new anti-HIV drugs.
Although D4TTP is not a good substrate for human DNA polymerases, its incorporation into DNA by pol a and pol c was detected in the DNA primer extension assay (Fig. 2). Thus, it is possible that this compound may be incorporated into cellular DNA in vioo, terminate chromosomal DNA replication, and ultimately lead to cytotoxicity. In contrast, no incorporation of AZTTP into DNA by human pol a and pol c was detectable by DNA primer extension assay in this and other studies (12). The cytotoxic effects of D4T and AZT have been reported both in cell culture and in mice (13,14,21,33). Because AZTTP is not a substrate for replicative human DNA polymerases, its inhibitory activity on DNA polymerase y, a mitochondrial enzyme, may be in part responsible for its cytotoxicity (34). Furthermore, induction of alteration of cellular dNTP pools (35) and inhibition of DNA repair by AZT (36,37) may also lead to cellular toxicity. pol c is a highly processive enzyme with 3' + 5'-exonuclease activity (38,39). This enzyme is the mammalian homologue of pol I1 in yeast (40) and is thought to be involved in both DNA replication (40) and repair (41,42). The present study demonstrated that pol t was able to effectively remove nucleotides from a normal 17-base DNA primer, but was unable to excise D4TMP from the 3'-end of an oligomer. AZTMP, however, was removed from the 3'-end of the DNA by pol e at about 20% of the rate for normal deoxynucleotide excision. Thus, it is likely that incorporation of D4TMP or AZTMP into the DNA primer by HIV-RT results in a steric change in the template-primer interaction that renders D4TMP impervious to and AZTMP less susceptible to excision by pol c. The inability of pol t to remove D4TMP from DNA may explain why the cytotoxic effect of D4T in H9 cells could not be reversed by addition of exogenous thymidine (33). Once the analogue is incorporated, it may remain in cellular DNA permanently. Whether DNA polymerase 6, another mammalian enzyme with 3' + 5'-exonuclease activity, can remove D4TMP from DNA is not known. Detailed investigations of the excision of these analogues by pol 6 and pol e in vitro and in whole cells will further our understanding of the drugs' actions.